68-AMERICA-45 in our Date fields

We're having a strange problem with some date fields in our database, and frankly we're not at all sure what's happening. Let me try to generally describe the symptoms, and maybe someone can help me figure out the problem in our system.

In short, we add and update all data in our database through stored procedures, and presently all external updates/adds occur through a Java servlet. This has been working fairly well for close to a year for various field types, but now we have some strange problem showing up.

We're have a situation in which some of our date fields (not many, but some) are getting set to nonsense dates such as "68-AMERICA-45" or
"83-AMERICA-49" or "79-AMERICA-52". As far as we know the servlet is
passing a proper dates to the date parameter in the stored procedure (it is a date parameter, not a varchar2). The procedure updates or adds the row to the table, and then when we look at it through SQL*Plus the field is sometimes set to one of the strange values above ("68-AMERICA-45" or
"83-AMERICA-49" or "79-AMERICA-52").

One of our database programmers claims to have witnessed this problem occurring repeatedly, then recompiled the procedure the servlet was using, and then the problem went away for a while. (I don't really know if this was strictly correllational or not, but I'll mention it anyway.)

It's difficult to repeatedly produce the problem, since some times for whatever reason it seems to work properly.

Perhaps someone can help point me in a direction to help me fix this problem. What would be a circumstance in which a date field might get set to "68-AMERICA-45" or "83-AMERICA-49" or "79-AMERICA-52"?

Thanks.

Jeff Braun
jeffb_at_halcyon.com

p.s.
I'm happy to repost any replies sent directly to me. Received on Wed Jan 05 2000 - 18:41:03 CST